HomeScienceData-Driven Model Identifies Structura
SCIENCE

Data-Driven Model Identifies Structural Basis of Amorphous Ices

Simulations published in Nature Communications reveal how local environments distinguish low-density and high-density amorphous ices during pressure-induced transitions.

WHAT YOU NEED TO KNOW
  • Nature Communications published the peer-reviewed research on Sept. 12, 2026.
  • Phase identity between LDA and HDA ice is encoded within the first coordination shell.
  • Local hydrogen density captured structural hysteresis during compression and decompression, whereas orientational order parameters did not.

Researchers from Princeton University and the City University of New York developed a probabilistic data-driven framework to isolate structural differences between amorphous ices, according to a study published in Nature Communications.

The team applied the method to molecular simulations of water. The calculations identified local collective variables that distinguish low-density amorphous (LDA) and high-density amorphous (HDA) ice. While local density descriptors separate the two states, the researchers found that phase identity is encoded inside the first coordination shell.

The pressure-induced transition between LDA and HDA proceeds through redistribution between LDA-like and HDA-like local environments. The simulations showed no evidence of intermediate structures, matching behavior typical of first-order phase transitions. This dynamic contrasts with the gradual structural changes observed in other non-crystalline materials, such as metallic glasses.

Local hydrogen density revealed pronounced structural hysteresis between compression and decompression pathways. Conventional orientational order parameters did not capture this hysteresis. The difference demonstrated that microscopic interpretations of amorphous transformations depend fundamentally on descriptor choice. The authors confirmed these results remained robust across multiple force fields.

Quinn M. Gallagher and Ryan J. Szukalo contributed equally as lead authors, collaborating with Nicolas Giovambattista, Pablo G. Debenedetti, and Michael A. Webb. The authors conducted the research across Princeton’s chemical and biological engineering and chemistry departments, Brooklyn College, and The Graduate Center of the City University of New York.

Princeton Research Computing supported the simulations through resources managed by the Princeton Institute for Computational Science and Engineering and the Office of Information Technology. Project funding came from the U.S. Department of Energy’s Chemistry in Solution and at Interfaces Center alongside several grants from the National Science Foundation.

Xentir Media
Xentir Media NewsroomSource-backed AI and technology coverage, drafted by Xentir's automated editorial system under fixed human-set rules. See our editorial policy and AI usage policy.
J
Jomon · Founder & EditorFounder and editor of Xentir Media. Sets the editorial rules the newsroom system runs under, and is accountable for its corrections. About Jomon · [email protected]
The Xentir Brief
The developments worth knowing — one useful email.
Get the Brief →